banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Separating overlapping click trains originating from multiple individuals in echolocation recordingsa)
a)This work was presented at the 5th Animal Sonar Symposium, Kyoto, Japan, 14–18 September 2009.
Rent this article for
View: Figures


Image of FIG. 1.
FIG. 1.

Typical measurement scenario resulting in overlapping click trains. (A) Hydrophone array with a sonar target in front. (B) Multiple free-swimming dolphins spontaneously echolocating concurrently at the same sonar target.

Image of FIG. 2.
FIG. 2.

(Color online) (A) Frequency spectra of successive clicks captured with hydrophone No. 17. (B) Spectrogram created by concatenating the frequency spectra from each click in (A).

Image of FIG. 3.
FIG. 3.

(Color online) (A) Peak frequency from one of the hydrophones (No. 17), ensonified throughout an echolocation sequence, where more than one animal were present. (B) Weighted mean frequency measured with the same hydrophone as in (A). (C) ICI intervals during the sequence. (D) Summarized measured energies across the entire hydrophone array (47 elements) throughout the sequence. (E) Correlation coefficients of successive frequency spectra throughout the sequence recorded by hydrophone No. 17.

Image of FIG. 4.
FIG. 4.

(Color online) Relative energies measured across the array (circles) plotted together with the occurrences of correlation coefficients below 0.95 (crosses). The latter have been plotted at the mean relative energy level of the two adjacent clicks for the sake of comparison.

Image of FIG. 5.
FIG. 5.

Result from automatic click train separation algorithm, obtained through the correlation results.

Image of FIG. 6.
FIG. 6.

(Color online) The click spectrograms individually grouped and concatenated sequentially according to the automatic click train grouping results. Click numbers refer to the sequential number of the clicks for each click train, i.e., not to their real time occurrence (cf. Fig. 5).

Image of FIG. 7.
FIG. 7.

Notched box and whisker plot showing the correlation coefficients for different click lags, i.e., the correlation between clicks 1 and 2, 2 and 3, 3 and 4 etc. (j = 1), clicks 1 and 3, 2 and 4, 3 and 5 (j = 2), etc. for j = 1, 2, 3, … , 50 [see Eq. (1)].

Image of FIG. 8.
FIG. 8.

(Color online) (A) Peak frequencies. (B) Centroid frequencies. (C) Relative energies across the array in each click during the sequence. (D) ICI intervals in the click trains; color and marker shape coded based on the frequency spectra correlation results.

Image of FIG. 9.
FIG. 9.

(A) Time signal with high SNR. (B) Frequency spectra of signal in (A). (C) Time signal with low SNR. (D) Frequency spectra of signal in (C).

Image of FIG. 10.
FIG. 10.

Notched box and whisker plots over the number of separated click trains for correlation coefficient thresholds 0.93, 0.95, and 0.97 and increasing signal noise levels. The noise levels were varied with respect to the original noise level by adding white Gaussian noise to the recorded signals.


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Separating overlapping click trains originating from multiple individuals in echolocation recordingsa)